As the classification process is initiated to convert remotely sensed data into meaningful information, such as an accurate mapping of complex land use/land cover types,a qualified classification method is required.In practice, a one-against-one (OAO) classification strategy, which divides a multi-class problem into sub-binaryproblems, has been widely used.However, the combination mechanisms currently used in OAO fail to separate non-competent classifiers from those have been trained by real class training samples.Thus, we introduced a distance-based Relative Competence Weighting (DRCW) mechanism into the combination process.Moreover, based on the support vector machine (SVM)algorithm, we developed a SVM-DRCW classification method for images classification.Using Guangzhou as a study area, the proposed technique wasapplied in classification of Landsat TM and SPOT HRG remote sensing images with moderate spatial resolution.Theresults show that SVM+DRCW can achieve higher classification accuracycompared to the one-against-all SVM and SVMs combined with weight voting or pairwise coupling strategies.